G github.com Show HN: We compressed a 31B video+audio diffusion model to 14B, 6 steps, INT8 83 points jiant 4 days ago 14 comments
